Journal of Pragmatics xxx (2010) xxx–xxx

PRAGMA-3162; No of Pages 8

Contents lists available at ScienceDirect

Journal of Pragmatics

journa l homepage: www.e lsev ier .com/ locate /pragma

Prosodic stress on a word directs 24-month-olds’ attention to acontextually new referent

Susanne Grassmann *, Michael Tomasello

Max-Planck-Institute for Evolutionary Anthropology, Deutscher Platz 6, 04103 Leipzig, Germany

A R T I C L E I N F O

Article history:

Received 7 July 2009

Received in revised form 30 December 2009

Accepted 11 April 2010

Keywords:

Language acquisition

Children

Contrastive stress

Information structure

Visual attention

A B S T R A C T

From the very beginning of language acquisition, young children are sensitive to what is

given versus what is new in their discourse with others. Here we ask whether 24-month-

olds use this skill to interpret prosodic highlighting as an invitation to focus their attention

on what is new in the situation. Using an eye-tracking methodology, we compared

children’s visual fixation of referents that were given versus those that were new in the

situation when the prosodic highlighting of their corresponding word varied. Results

showed that 24-month-old children looked longer to the referents of prosodically stressed

words when those referents were new to the context. Neither stress of the word alone nor

newness of the referent alone was sufficient to induce children to focus their attention on

the target referent. These results suggest that from an early age children understand at

least one important communicative function of prosodic stress.

� 2010 Elsevier B.V. All rights reserved.

1. Introduction

From an early age children are sensitive to the distinction between new and given information and tailor theircommunicative acts accordingly (e.g., Greenfield, 1979; ONeill and Happé, 2000). Children furthermore use the distinctionbetween given and new information when interpreting other peoples’ attention (e.g., Moll and Tomasello, 2007; Moll et al.,2006) andwhen learning novel words (e.g., Akhtar et al., 1996; Tomasello and Akhtar, 1995). In Tomasello and Akhtar’s (1995)study, young 2-year-olds heard an adult say a novel wordwhen they jointly looked at a nameless object that was engaged in anameless action. In one condition, the discourse situation preceding the naming event was manipulated so that the namelessaction was the new element when the novel word was said. In another condition it was the nameless object that was the newelement. Tomasello and Akhtar found that children learned the novel word forwhatever the new element was in the situationwhen the novelwordwas said. They suggest, followingGreenfield (1979), that discourse newness is a powerful factor inducingjoint attention and that children know this and use it in their interpretation of their interlocutor’s referential intention.

In a recent word learning study however, Grassmann and Tomasello (2007) demonstrated that young 2-year-olds onlyrelied on discourse newness in their interpretation of a novel word, when the novel word was accentuated.1 In one of theconditions in this study a nameless novel object was new to the situation and in another condition a nameless action wasnew to the naming situation – as in Tomasello and Akhtar (1995). However, in Grassmann and Tomasello’s study, childrenheard the experimenter say two novel words in an intransitive sentence, a novel verb (miekt) as well as a novel noun (feks):Der Feks miekt. As a second factor, sentence accent was varied: either the novel noun or the novel verb was accentuated. The

* Corresponding author. Tel.: +49 341 3550 341; fax: +49 341 3550 444.

E-mail address: grassman@eva.mpg.de (S. Grassmann).1 The terms ‘‘stress’’ and ‘‘accentuation’’ are used synonymously throughout this text for acoustic saliency on the utterance level.

Please cite this article in press as: Grassmann, S., Tomasello, M., Prosodic stress on a word directs 24-month-olds’attention to a contextually new referent, Journal of Pragmatics (2010), doi:10.1016/j.pragma.2010.04.019

0378-2166/$ – see front matter � 2010 Elsevier B.V. All rights reserved.

doi:10.1016/j.pragma.2010.04.019

S. Grassmann, M. Tomasello / Journal of Pragmatics xxx (2010) xxx–xxx2

PRAGMA-3162; No of Pages 8

results revealed that children learned the novel noun (Feks) for the novel object onlywhen the nounwas accentuated and thenovel object was new in the situation but not when the noun was accentuated and the novel object given. Grassmann andTomasello (2007) suggest that this indicates that children interpret sentence accent in language as being iconic of thespeaker’s intention to refer to a salient aspect of the situation.

It is an important communicative function of language to direct others’ attention and establish continuously changingjoint attentional foci. For example, when a speaker wants to establish shared attention towards a dog in his vicinity, he couldjust say the word dog. However, usually verbal communication consists of more than one word per utterance. That is, ratherthan saying just one word, the speaker might say a whole sentence such as The dog is playing with the ball. With such asentence the speaker could express the intention to jointly focus on the situation as a whole, or alternatively he/she mightwant to focus particularly on one of the participants (e.g., the dog) or even the action. To highlight the piece of informationthat is intended as the focus of the listener’s attention, speakers of intonational languages, such as German and English, useprosodic stress (e.g., Bolinger, 1961; Chafe, 1972, 1994; Lambrecht, 1994).

Since an individual’s visual attention follows the speech they hear (Cooper, 1974; Golinkoff et al., 1987; Swingley et al.,1999; for a review see Tanenhaus and Trueswell, 2005) one possibility for investigating the attentional effects of prosodicstress would be to use a looking-time method. To date, there is just one study that we are aware of (Arnold, 2008) that hasused such a method to examine children’s comprehension of stress. But, although Arnold (2008) demonstrated that 4–5-year-olds assume that unstressed words refer to previously established referents, no evidence was found in this study tosuggest that children interpret stressed words as referring to new information. Other studies of 3–14-year-olds’comprehension of sentence stress have used alternative methods and produced mixed results. The youngest age for whichone of these studies demonstrated children’s knowledge of the newness-stressed and givenness-unstressed relation was 6years of age (Cutler and Swinney, 1987; Hornby, 1971). However, MacWhinney and Price (1980) used the same method asHornby (1971) but found comprehension only in 14-year-olds. These findings are surprising given that there is evidence ofchildren’s early mastery of the stress-newness relation from production studies starting as young as 2–3 years of age (e.g.,Baltaxe, 1984; Hornby and Hass, 1970; MacWhinney and Bates, 1978; Pechmann, 1981; Wieman, 1975), and given thatGrassmann and Tomasello (2007) found that even 25-month-olds use the stress–newness relation in word learning.

The question is thus,what is children’s early understanding of the function of stress? The comprehension studies from the1970s and 1980s only investigated the role of stress in children’s interpretation of the information structure of sentences,that is ‘‘what is the sentence about’’ or ‘‘what contrasts the sentence with from a logical perspective?’’ However, althoughyoung children do not fully understand such abstract concepts that are related to prosodic stress, it might be – as suggestedby Grassmann and Tomasello (2007) – that young children do understand the discourse function of stress as a means todirect a listener’s attention to a new discourse referent. In order to support this view, it must be shown that children orientand focus their visual attention to a new discourse referent when a speaker is stressing a corresponding word in his or herutterance. Furthermore, if the hypothesis is correct, children should not focus their attention to the new discourse referentwhen the speaker is not intending to direct their attention to it (and thus does not stress a corresponding word). In this casechildrenmight interpret the speaker’s intention to refer to a situation as a whole with no specific individual element in focusand it would then be up to the next conversational turn to establish which discourse referent will be in the joined focus ofattention.

In the current study we investigated 24-month-olds comprehension of prosodic stress using a looking-timemeasurement. We asked whether children focus their visual attention on new referents when the corresponding wordis stressed in an utterance. First, we showed the children pictures of highly familiar objects (e.g., a ball). Then we showedthem a second picture with two highly familiar objects, one of which was given from the first picture and the other, whichwas new. Before the second picture was revealed, the children heard a sentence such as The dog has a ball – where the stressfell either on dog or on ball. We expected that the children would focus their attention on the new referents of the stressedwords. The effect of newness and stress on children’s visual attention in isolation was tested in two control conditions.

2. Method

2.1. Participants

Thirty-six monolingual German-speaking children aged �2 weeks around their second birthdays participated in the study(range 1;11,16 to 2;0,13, mean = 1;11,29). Their parents had previously volunteered to participate in child development studies.Six additional children participated butwere dropped from the study because of the experimenter’s error (2), technical failure (1),disinterest in the pictures (2), or the non-visibility of their eyes in the video (1). Children were randomly assigned to one of theconditions, so that there was a total of 12 children in each condition. The number of boys and girls in each condition wasapproximately 50%.

2.2. Materials and design

We showed the children colored line drawings (approx. 5 cm � 5 cm) of highly familiar objects. Four pairs of pictureswere chosen so that their corresponding words had a similar syllabic structure but maximally distinct phonological forms:dog – ball (hund, ball), cat – car (katze, auto), teddy bear – flower (teddy, blume) and duck – apple (ente, apfel).

Please cite this article in press as: Grassmann, S., Tomasello, M., Prosodic stress on a word directs 24-month-olds’attention to a contextually new referent, Journal of Pragmatics (2010), doi:10.1016/j.pragma.2010.04.019

S. Grassmann, M. Tomasello / Journal of Pragmatics xxx (2010) xxx–xxx 3

PRAGMA-3162; No of Pages 8

These pictures were presented to children on a 19 in. computer monitor. Verbal descriptions of the pictures werespoken live. The children first saw a context picture and then a target picture (see Fig. 1). Children were randomlyassigned to one of three between-subjects conditions. In all conditions, the target picture showed both picturesbelonging to the pair. The context picture showed only one element of the pair in the Newness & Stress Condition andNewness Only Condition. Therefore, in these conditions one element in the target picture was new information and inthe other it was old. In the Stress Only Condition both elements in the target picture had already been presented in thecontext picture; thus old information was presented in the target picture. In all conditions the target picture wasdescribed verbally (e.g., The dog has the ball). The three conditions differed with respect to the locus of stress in thissentence. In the Newness & Stress Condition the word corresponding to the new element in the target picture wasstressed. The Stress Only Condition used the same stress pattern as that used in the Newness & Stress Condition. In theNewness Only Condition the sentence was produced without a saliently stressed element. Stress was realized as LH*accent. Furthermore, the intensity and duration of a stressed word was increased compared to the same word when itwas unstressed (see Fig. 2).

Since we were interested in children’s knowledge of the function of stress in communication, we assessed children’slooking behavior in a quasi-realistic setting with the experimenter talking live to the child. In order to ensure that allutterances had the appropriate prosodic characteristics, we analyzed recordings of all target utterance. A coder judged foreach utterance whether or not it contained a stressed element and if yes, which element was the stressed one. In theNewness Only Condition the experimenter misspoke in six cases and stressed the new element. The corresponding trialswere excluded from analyses.

Eachpicture-pairwas presented as a target screen inone trial; thus childrenparticipated in a total of four trials. In the targetpicture the pair of pictures was vertically centered and each element was 2 cm from the horizontal edge (thus, a 20–24 cm

Fig. 1. Procedures in the three experimental conditions. Newness & Stress Condition (top), Stress Only Condition (middle), and Newness Only Condition

(bottom).

Please cite this article in press as: Grassmann, S., Tomasello, M., Prosodic stress on a word directs 24-month-olds’attention to a contextually new referent, Journal of Pragmatics (2010), doi:10.1016/j.pragma.2010.04.019

Fig. 2. Example for the intonation of the target picture description in the Newness & Stress Condition and the Stress Only Condition (top) and in the Newness

Only Condition (bottom). Blue lines (non-continuous) represent the pitch contours, green lines (continuous) represent intensity. (For interpretation of the

references to color in this figure legend, the reader is referred to the web version of the article.)

S. Grassmann, M. Tomasello / Journal of Pragmatics xxx (2010) xxx–xxx4

PRAGMA-3162; No of Pages 8

spacing between the two pictures). The left/right position of the pictures, as well as whether the referent of the subject or theobject in the sentencewas the target element,was counterbalancedwithinandacross children. The target screenwasvisible for6 s. The children’s fixations were recordedwith a video camera (resolution 40ms) while theywere looking towards the targetscreen.

2.3. Procedure

The children visited the lab for approximately 15 min. During the study they sat on the parent’s lap at a table(70 cm � 70 cm), with the experimenter (E) on their left hand side. The computer monitor was positioned on the tableapproximately 50 cm in front of the child. All sessions were videotapedwith a camera centered behind the screen in order torecord the child’s eye movements. Parents had their eyes closed during the experimental trials. Before the experimentalsession, parents confirmed that their child knew the words for the pictures that were used in the study.

2.3.1. Familiarization

The rationale behind this phasewas to familiarize the childrenwith the computermonitor and the study’s procedure. Thefirst three pictures displayed on themonitorwere colored line drawings of 12 familiar objects. The childrenwere encouragedto name them or point to them. All 8 pictures that would be used in the experimental trials were among the 36 pictureschildren saw in this phase. The rationale behind this was to ensure that the children had no problems identifying thesepictures. The child then saw three further pictures that were structurally similar to the pictures that would be shown in thestudy. First the child saw a picture of a frog for 4 s and E said, Look a frog. Look the Frog. Then the child saw a picture with amoon and a pen for 4 s and E said, Look, the moon has the pen. The moon has the pen. Each picture started with a 100 ms soundto attract the child’s attention towards the monitor.

Please cite this article in press as: Grassmann, S., Tomasello, M., Prosodic stress on a word directs 24-month-olds’attention to a contextually new referent, Journal of Pragmatics (2010), doi:10.1016/j.pragma.2010.04.019

S. Grassmann, M. Tomasello / Journal of Pragmatics xxx (2010) xxx–xxx 5

PRAGMA-3162; No of Pages 8

2.3.2. Experimental trials

The procedure of the experimental trials in all three conditions is depicted in Fig. 1.Newness & Stress Condition. The children first saw a context picturewith one element centered on themonitor. E labeled

the context picture twice,Guckmal, da ist ein Hund. Guckmal, der Hund. (Look, there is a dog, look, the dog.). After 3 s, the contextpicturedisappeared and left a blank screen. ThenE said for example,DerHundhat den BALL (The dog has the BALL.) –wherein thenew element was stressed (see Fig. 2 for the prosodic details) – and then the target picture was presented for 6 s. When thescreen went blank again, E told the child that she wanted to show him/her another picture and the next trial started.

Newness Only Condition. The procedure of the Newness Only condition differed to the Newness & Stress Condition onlyin respect to stress. In this condition, both nouns were similarly stressed in the verbal description of the target screen (seeFig. 2 for prosodic details).

Stress Only Condition. The difference between the Stress Only Condition and the Newness & Stress Condition wasthat in the Stress Only Condition both elements in the target picture were old information: the target picture showedthe same pair of pictures that were visible in the context picture – in a different orientation. E named both pictures inthe context picture. Top–bottom position and order of labeling were counterbalanced within and across children. In theStress Only Condition the prosodic characteristics of the target picture’s description were identical to the Newness &Stress Condition.

2.4. Coding and reliability

Children’s looking was coded frame by frame from Videotape as ‘‘left’’, ‘‘right’’, or ‘‘undetermined’’. Ten children each hadone trial excluded either because they did not know the label of an object used in that particular trial or because E misspoke(4 children in the Newness & Stress Condition, 6 children in the Newness Only Condition, and 3 children in the Stress OnlyCondition). A second independent coder scored fixations of nine randomly selected children (three in each condition). Asestimated by Cohen’s Kappa frame-wise, inter-observer reliability was 0.93.

During the running of the experiment we noticed that children frequently pointed to either of the two elements in thetarget screen.We codedwhether these pointing gestureswere directed to the left or the right picture (which according to thecounterbalancing corresponded to the target or non-target picture). When children pointed to both pictures successively,only the first pointing gesture was counted. A second independent coder scored a random selection of nine children (three ineach condition). The two coders agreed 100%.

3. Results

3.1. Looking

Preliminary analyses revealed no effect of target position (left/right) and also no effect of sentence position (first/secondargument). First, we compared the mean proportions2 of the children’s looking behavior to chance (50%) to test how oftenthe children’s attention was focused towards the target in each condition. Table 1 shows the mean proportions of children’slooks towards the target during the first 2 s aswell as thewhole 6 s that the target screenwas presented. Only in theNewness& Stress Condition did the children lookmore at the target picture (the new element in the screen whose namewas stressedin the verbal description of the screen) than was expected by chance (first 2 s: t(11) = 2.905, p = 0.014, whole 6 s: t(11)= 3.326, p = 0.004, one-sample t-tests). In the other two conditions the children looked towards the target at chance level.That is, in the Newness Only Condition they looked towards the new element at chance level – thus they spent an equalamount of time looking towards the given and the new element – whose labels were equal in prosodic salience. In the StressOnly Condition children looked equally towards the referents of the stressed and the unstressed word – which were equallygiven from the preceding context.

Analyses of the individual children’s looking behavior revealed the same pattern. Each child in the Newness & StressCondition looked on average more to the target picture than to the non-target picture. In contrast, in the other twoconditions only 7 out of 12 children looked on averagemore to the target picture and 5 children looked on averagemore tothe non-target picture (x2 = 6.92, df = 2, p = 0.016, one-sided Fisher exact test). However, the individual children’s analysisalso indicates that although the children in the Newness & Stress Condition focused their attention towards one of the twopictures in nearly all of their trials, itwasnot always the target picture–butmore thanwouldhave beenexpectedby chance(M = 68.75, SD = 17.81, t(11) = 3.647, p = 0.0038). In the Newness Only Condition and in the Stress Only Condition childrenfocused the target picture in on average half of the trials (2.08 trials both conditions, mean proportions of trials looking totarget: Newness Only Condition, M = 54.17, SD = 27.18; Stress Only Condition, M = 56.25, SD = 17.81). Thus, in these twoconditions, children’s looking behavior was presumably only driven by their individual preferences for one of the picturesin each trial.

Thus, the analyses of the average proportions children looked at the target picture revealed that when a stressed wordreferred to a contextually new picture, the children focused their attention towards it. But neither newness nor prosodic

2 Denominator = time-frame duration minus time coded as ‘‘undetermined’’.

Please cite this article in press as: Grassmann, S., Tomasello, M., Prosodic stress on a word directs 24-month-olds’attention to a contextually new referent, Journal of Pragmatics (2010), doi:10.1016/j.pragma.2010.04.019

Table 2Mean duration inmilliseconds (and standard deviations) of the children’s looks towards the target and non-target picture

in each condition.

Total looking times in ms

Target Non-target Total

Newness & Stress Condition 2656 (SD = 555) 1866 (SD = 388) 4522 (SD = 593)

Newness Only Condition 2578 (SD = 523) 2245 (SD = 620) 4823 (SD = 528)

Stress Only Condition 2201 (SD = 543) 1969 (SD = 423) 4169 (SD = 647)

Table 1Mean proportions (and standard deviations) of the children’s looks towards the target picture in each condition in the first

2 s as well as during the whole 6 s the target screen was visible.

2 s 6 s

Newness & Stress Condition 0.60* (SD = 0.08) 0.59* (SD = 0.08)

Newness Only Condition 0.48 (SD = 0.10) 0.54 (SD = 0.11)

Stress Only Condition 0.52 (SD = 0.09) 0.52 (SD = 0.09)

* p < 0.05 (one-sample t-test).

S. Grassmann, M. Tomasello / Journal of Pragmatics xxx (2010) xxx–xxx6

PRAGMA-3162; No of Pages 8

salience were sufficient in order to focus the children’s attention on the corresponding referent in the situation and tooverride individual preferences.

In a further analysis we compared the children’s total looking times in the three conditions. The rationale behind doingthiswas that the children’s overall interest in the target screen as awhole, and in each of the two elements,might be differentfor the three conditions independently of how frequently they looked proportionally towards the target element (see above).Table 2 shows the mean duration of time that the children looked towards the target screen and each of the depictedelements. The results show that the children had more overall interest in the target screen and looked longer at it in the twoconditions inwhich a new elementwas depicted (Newness & Stress Condition: 4522 ms, Newness Only Condition: 4823 ms)than in the Stress Only Condition (4169 ms) in which no new element was presented (F(2,33) = 3.664, p = 0.037, ƞ2 = 0.182one-sided LSD posthoc comparisons: Newness & Stress Condition vs. Stress Only Condition, p = 0.011, Newness OnlyCondition vs. Stress Only Condition, p = 0.011).

To summarize, we found that although scenes that contain a new element elicit more interest than scenes that onlycontain given elements, only prosodic highlighting of the corresponding verbal expression focuses a child’s attentiontowards the new element.

3.2. Pointing

Table 3 shows themean proportion of trials in which the children pointed to any of the two elements in the target screenin each condition, aswell as themean proportion of points directed to the target element. Overall, when the children pointed,theyweremore likely to point to the target than to the non-target element – that is, to the new element in the Newness OnlyCondition, to the stressed element in the Stress Only Condition and to the new and stressed element in the Newness & StressCondition (M = 67%, SD = 27). A one-way ANOVA revealed no difference between conditions.

The three conditions differed however with respect to how frequent pointing was in each condition. Pointing was mostfrequent in the Newness Only Condition (M = 73%, SD = 27). In the other two conditions children pointed significantly less:Newness & Stress Condition, 49%, Stress Only Condition, 34%, F(2,33) = 3.843, p = 0.032, ƞ2 = 0.189, one-sided LSD posthocTests: Newness Only vs. Newness & Stress: p = 0.047, Newness only vs. Stress Only: p = 0.005. An analysis of the number ofchildren who pointed in at least one trial revealed the same results: all children in the Newness Condition, in which noelement was stressed, pointed at least once (N = 12). In contrast, in both conditions with a stressed word (Newness & Stress

Table 3Mean proportions of trials in which the children pointed (left) and mean proportions from those trials with pointing in

which children pointed to the target picture in each condition (right).

Mean proportions of

Trials with pointing Pointing to target*

Newness & Stress Condition 0.49 (SD = 0.43) 0.62 (SD = 0.18)

Newness Only Condition 0.73 (SD = 0.27) 0.66 (SD = 0.28)

Stress Only Condition 0.34 (SD = 0.32) 0.73 (SD = 0.36)

* Denominator = number of trials with pointing per child (chance level = 0.5).

Please cite this article in press as: Grassmann, S., Tomasello, M., Prosodic stress on a word directs 24-month-olds’attention to a contextually new referent, Journal of Pragmatics (2010), doi:10.1016/j.pragma.2010.04.019

S. Grassmann, M. Tomasello / Journal of Pragmatics xxx (2010) xxx–xxx 7

PRAGMA-3162; No of Pages 8

Condition and Stress Only Condition) fewer children pointed (N = 8 in both conditions,x2 = 5.143, df = 2, p = 0.032, one-sidedFisher exact test).

To summarize, children pointedmost often in the Newness Only Condition, whichwas the only condition in which stresswas not applied. In other words: children were less likely to point to an element in a scene when one element had beenprosodically highlighted in the verbal description of that scene.

4. Discussion

In the current study we examined whether a combination of sentence accent on a word and contextual newness of thecorresponding referent led 24-month-olds to focus their visual attention on that referent. Results show that children didindeed focus their visual attention on the referent of a familiar word when the word was accentuated and the referent wasnew to the situation. Importantly, neither accentuation on aword nor newness of a referent alone led the children to visuallyfocus on the corresponding element (i.e., the referent of the acoustically salient word or the new element in the situation).This suggests that children assume that the acoustic saliency of words is iconic of the contextual saliency of the referents –perhaps because this has been conventionalized by the language they are learning as expressing the speaker’s intention ofestablishing a new joint focus of attention with the listener on something that is new in a situation.

Although linguists have speculated that the function of prosodic stress is to establish a new focus of attention (e.g.,Bolinger, 1961; Chafe, 1994; Lambrecht, 1994), the current study is the first show that empirically and with young children.That is, the current study shows that by the time children start producing multi-word utterances and use stress to highlightnew information therein (Wieman, 1975), they also understand the communicative function of stress in languagecomprehension. This is an important finding since previous research has often proposed asymmetries between children’scomprehension and production of prosodic stress, such that production seems to precede comprehension (e.g., Cutler andSwinney, 1987; Gualmini et al., 2003). On the one hand, production studies have shown that children consistently stress newand contrastive information from 2 to 3 years onwards (Baltaxe, 1984; Hornby and Hass, 1970; MacWhinney and Bates,1978; Pechmann, 1981; Wieman, 1975). Comprehension studies, on the other hand, have found that only by the age of 6years (or even later) children interpret stress correctly and know about the relation between stress and new information(Cutler and Swinney, 1987; Gualmini et al., 2003; Hornby, 1971; MacWhinney and Price, 1980). However, these earlierproduction and comprehension studies employ methods of differing complexity. Specifically, the production studies usedsimple tasks such as requesting objects one after another (Pechmann, 1981) or describing pictures successively (Baltaxe,1984; Hornby and Hass, 1970; MacWhinney and Bates, 1978) while earlier comprehension studies used more complexmethodologies, such as asking children about the meaning of a sentence (Hornby, 1971; MacWhinney and Price, 1980),truth-value-judgments (Gualmini et al., 2003), or phoneme detection (Cutler and Swinney, 1987). Thus, given that earliercomprehension studies used more complex methods than earlier production studies, findings of a delayed comprehensionmust be taken with caution. Themethod of the current study provides a much fairer comparison with the earlier productionstudies. Importantly, our findings do not support a delayed comprehension of children’s understanding of the function ofstress as marking new information.

The finding that children do not look longer at the target in the Stress Only condition may indirectly indicate that youngchildren interpret emphatic stress differently from stress that highlights new information. It is not clear what exactly thediscourse function of emphatic stress is from a theoretical perspective. Since children did not focus their attention towardsthe referent of a stressed word in the Stress Only Condition it seems likely that children only interpret stress asexpressing the speaker’s intention to jointly focus the referent of the stressed element when the speaker plausibly couldhave intended this (but not in the absence of any obvious reason for making an discourse referent more salient). Thisdoes however not mean that children do ignore the prosodic characteristics of the speaker’s utterance in the Stress OnlyCondition. Children’s pointing behavior in the current study is of interest in this respect. We found that children wereless likely to point in the Newness & Stress Condition and in the Stress Only condition than in the Newness OnlyCondition. This pattern suggests that pointing and stress have similar functions: namely, to establish a central elementin the discourse. In the Newness Only Condition the verbal description only focused very broadly on the whole scenebecause no particular element was stressed. One could interpret the children’s frequent pointing in this condition asindicative of their having missed a clear focus in the verbal description of the target picture and of their then having-established one in their next conversational turn.

The absence of pointing in the two conditions in which the speaker stressed one element thus may indicate that childreninterpreted stress as establishing the stressed element as central to the subsequent discourse.

The findings of the current study are also relevant in order to explain the results of Grassmann and Tomasello’s (2007)word learning study. In this study, children learned a novel noun that was presented together with a novel verb in anintransitive sentence (The feks is meeking.), but only when a novel object was new in the corresponding situation and thenoun was stressed in the utterance. As in the current study, neither newness nor stress had an effect in isolation. However,newness of one element was confounded with stress on the other element and stress on one word was confounded withnewness of the other element. Thus, there was no ‘‘newness only’’ and no ‘‘stress only’’ condition. The current studydisentangled these two factors. The current results are that children focus their attention to new discourse referents whenthe corresponding word in the utterance is stressed. This suggest the following explanation for children’s word learning inGrassmann and Tomasello (2007): it might be that when children hear a stressed word they assume that it refers to

Please cite this article in press as: Grassmann, S., Tomasello, M., Prosodic stress on a word directs 24-month-olds’attention to a contextually new referent, Journal of Pragmatics (2010), doi:10.1016/j.pragma.2010.04.019

S. Grassmann, M. Tomasello / Journal of Pragmatics xxx (2010) xxx–xxx8

PRAGMA-3162; No of Pages 8

something that is new in the situation. Since in Grassmann and Tomasello’s study there was always something new in thesituation, children might have learned the stressed word for whatever was new in the situation. That is, children’s not-learning of an unstressed novel noun for a discourse-new novel-object-referent might actually be due to learning thestressed novel word (in this condition the novel verb) for the discourse-new object. And vice versa, children’s not-learning ofa stressed novel noun for a discourse-given novel object might be due to learning the stressed novel noun for the discourse-new referent (in this case a novel action). The way in which children’s learning was assessed in Grassmann and Tomasello(2007) made it impossible to test for such cross-category learning since for testing noun learning only novel objects wereavailable and for testing verb learning only novel actions were available. Therefore, only when newness and stress occuredtogether were children able to find the correct referent of the novel noun.

In the current study accentuation was realized by several acoustic parameters: accent type, F0, intensity, duration.Although pitch variation and this is generally considered to be the primary cue to focus (Dahan and Bernard, 1996),accentuation generally also involves changes in duration, intensity and vowel quality (Turk and White, 1999; Xu and Xu,2005). The current study leaves open how much each of the prosodic parameters influences children’s response. Futureresearch needs to investigate the effect of each parameter in isolation.

To summarize, the current study demonstrates thatwhen 2-year-old children hear a stressedword in a verbal descriptionof a situation they search for a new referent in the situation and focus their visual attention on it. Our findings suggest thatchildren’s comprehension of prosodic stress develops around the same time as their early use of prosodic stress to highlightnew information.

References

Akhtar, N., Carpenter, M., Tomasello, M., 1996. The role of discourse novelty in early word learning. Child Development 67, 635–645.Arnold, J., 2008. THE BACON not the bacon: how children and adults understand accented and unaccented noun phrases. Cognition 108, 69–99.Baltaxe, C., 1984. Use of contrastive stress in normal, aphasic, and autistic children. Journal of Speech and Hearing Research 27, 97–105.Bolinger, D.L, 1961. Contrastive accent and contrastive stress. Language 37, 83–96.Chafe, W., 1972. Discourse structure and human knowledge. In: Roy, O., Freedle, John, B., Carroll, (Eds.), Language Comprehension and the Acquisition of

Knowledge. pp. 41–69.Chafe, W., 1994. Discourse, Consciousness, and Time. University of Chicago Press, Chicago.Cooper, R., 1974. The control of eye fixation by the meaning of spoken language. Cognitive Psychology 6, 84–107.Cutler, A., Swinney, D., 1987. Prosody and the development of comprehension. Journal of Child Language 14, 145–167.Dahan, D., Bernard, J.-M., 1996. Interspeaker variability in emphatic accent production in French. Language & Speech 39, 341–374.Golinkoff, R., Hirsh-Pasek, K., Cauley, K., 1987. The eyes have it—lexical and syntactic comprehension in a new paradigm. Journal of Child Language 14, 23–45.Grassmann, S., Tomasello, M., 2007. Two-year-olds use primary sentence accent to learn new words. Journal of Child Language 34, 677–687.Greenfield, P., 1979. Informativeness, presupposition, and semantic choice in single-word utterances. In: Ochs, E., Schieffelin, B.B. (Eds.), Developmental

Pragmatics. Academic, New York, pp. 159–166.Gualmini, A.,Maciukaite, S., Crain, S., 2003. Children’s insensitivity to contrastive stress in sentenceswithONLY. In: Proceedings of the 25th Penn Linguistics

Colloquium. pp. 87–110.Hornby, P., 1971. Surface-structure and topic-comment distinction—developmental study. Child Development 42, 1975–1988.Hornby, P.A., Hass, W.A., 1970. Use of contrastive stress by preschool children. Journal of Speech and Hearing Research 13, 387–394.Lambrecht, K., 1994. Information Structure and Sentence Form: Topic, Focus, and the Mental Representation of Discourse Referents. Cambridge University

Press.MacWhinney, B., Bates, E., 1978. Sentential devices for conveying givenness and newness: a cross-cultural developmental study. Journal of Verbal Learning

and Verbal Behaviour 17, 539–558.MacWhinney, B., Price, D., 1980. The development of the comprehension of topic-commentmarking. In: Ingram, D., Peng, C.C., Dale, P. (Eds.), Proceedings of

the First International Congress for the Study of Child Language. University Press of America, Lanham.Moll, H., Koring, C., Carpenter, M., Tomasello, M., 2006. Infants determine others’ focus of attention by pragmatics and exclusion. Journal of Cognition and

Development 7, 411–430.Moll, H., Tomasello, M., 2007. How 14- and 18-month-olds know what’s new for others. Developmental Psychology 43, 309–317.ONeill, D., Happé, F., 2000. Noticing and commenting on what’s new: differences and similarities among 22-month-old typically developing children,

children with Down syndrome and children with autism. Developmental Science 3, 457–478.Pechmann, Th., 1981. Children’s assignment of acoustic stress in referential communication. Papers and Reports on Child Language Development 20, 100–

107.Swingley, D., Pinto, J., Fernald, A., 1999. Continuous processing in word recognition at 24 months. Cognition 71, 73–108.Tanenhaus, Trueswell, 2005. Eye movements as a tool for bridging the language-as-product and language-as-action traditions. In: Trueswell, Tanenhaus,

(Eds.),Approaches to StudyingWorld-Situated Language Use: Bridging the Language-as-Product and Language-as-Action Traditions. MIT Press, pp. 3–37.Tomasello, M., Akhtar, N., 1995. Two-year-olds use pragmatic cues to differentiate reference to objects and actions. Cognitive Development 10, 201–224.Turk, A., White, L., 1999. Structural influences on accentual lenghtening. Journal of Phonetics 27, 171–206.Wieman, L., 1975. Stress patterns of early child language. Journal of Child Language 3, 283–286.Xu, Y., Xu, C.X., 2005. Phonetic realization of focus in English declarative intonation. Journal of Phonetics 33, 159–197.

S. Grassmann is a postdoctoral affiliate at the Department of Developmental and Comparative Psychology at the Max Planck Institute for EvolutionaryAnthropology in Leipzig, Germany. Her research interests are the development of children’s understanding of prosodic information structure marking and thepragmatic nature of the processes underlying children’s word learning. Publications include Two-year-olds exclude novel objects as potential referents of novelwords based on pragmatics (Cognition, 2009), Young children follow pointing over words in interpreting acts of reference (Developmental Science, 2009), Two-year-olds use primary sentence accent to learn new words (Journal of Child Language, 2007).

M. Tomasello taught at Emory University and worked at Yerkes Primate Center (USA) from 1980 to 1998; since 1998, Co-Director, Max Planck Institute forEvolutionary Anthropology, Leipzig, Germany. Research interests focus on processes of social cognition, social learning, and communication/language in humanchildren and great apes. Books include Primate Cognition (Oxford University Press, 1997), The New Psychology of Language: Cognitive and Functional Approachesto Language Structure (vols. 1 and 2; edited, Erlbaum, 1998; 2002), The Cultural Origins of Human Cognition (Harvard University Press, 1999), Constructing aLanguage: A Usage-Based Theory of Language Acquisition (Harvard U. Press, 2003), and Origins of Human Communication (MIT Press, 2008).

Please cite this article in press as: Grassmann, S., Tomasello, M., Prosodic stress on a word directs 24-month-olds’attention to a contextually new referent, Journal of Pragmatics (2010), doi:10.1016/j.pragma.2010.04.019